untitled design

Yandex speech recognition service taught to punctuate

Today, April 20, the developers of the Yandex Cloud cloud platform announced an important update to the Yandex SpeechKit service, built on the basis of machine learning. This technology is designed for speech recognition and synthesis, and now, when transforming voice into text, the cloud service has been taught to automatically place punctuation marks in the right places. The innovation is called “Punctuator” – it works both when recognizing the user’s speech in real time (for example, in scenarios with voice assistants), and when transforming speech into text when processing ready-made audio files.

To implement Punctuator, developers from Yandex had to use two machine learning models at once – the first one is responsible for transforming voice into text, and the second one compares the finished text with the rules and norms of the Russian language, after which it places the appropriate punctuation marks. It is worth noting that so far we are talking about punctuation marks exclusively for the Russian language – the developers did not specify when other languages ​​\u200b\u200bwould appear, but stated that the Punctuator is available to all Yandex SpeechKit users. And, of course, in the future it will give noticeable results.

The company’s specialists said that the innovation should improve the user experience in scenarios when it comes to direct interaction with a person. For example, a voice assistant will better perceive and understand human speech, and the technology of automatic transcription and subtitle formation for videos will improve.

Source: Trash Box

You may also like

Get the latest

Stay Informed: Get the Latest Updates and Insights

 

Most popular